Influence of early gut microbiota on the
maturation of childhood mucosal and systemic
Y M Sjögren, Sara Tomicic, Anna Lundberg, Malin Fagerås-Böttcher, B Björkstén, E
Sverremark-Ekström and Maria Jenmalm
Linköping University Post Print
N.B.: When citing this work, cite the original article.
This is the pre-reviewed version of the following article:
Y M Sjögren, Sara Tomicic, Anna Lundberg, Malin Fagerås-Böttcher, B Björkstén, E
Sverremark-Ekström and Maria Jenmalm, Influence of early gut microbiota on the maturation
of childhood mucosal and systemic immune responses, 2009, Clinical and Experimental
Allergy, (39), 12, 1842-1851.
which has been published in final form at:
Postprint available at: Linköping University Electronic Press
Influence of early gut microbiota on the maturation of childhood
mucosal and systemic immune responses.
Gut flora and immune responses.
Ylva M. Sjögren1, Sara Tomicic2, Anna Lundberg2, Malin F. Böttcher2, Bengt
Björkstén3, Eva Sverremark-Ekström1, Maria C. Jenmalm2
1The Department of Immunology, the Wenner-Gren Institute, Stockholm University,
2 The Division of Paediatrics, the Department of Clinical and Experimental Medicine, Faculty
of Health Sciences, Linköping University, Linköping, Sweden.
3 The Institute of Environmental Medicine, Karolinska Institutet, Stockholm, Sweden.
Ylva M. Sjögren
The Department of Immunology
The Wenner-Gren Institute
Arrhenius laboratory of Natural Sciences F5
Svante Arrhenius väg 16-18
106 91 Stockholm
Phone: +46 8 16 44 36
Fax: +46 8 6129542
Introduction: Among sensitized infants those with high, as compared with low levels, of
salivary secretory IgA (SIgA) are less likely to develop allergic symptoms. Also, early
colonization with certain gut microbiota, e.g. Lactobacilli and Bifidobacterium species, might
be associated with less allergy development. Although animal and in vitro studies emphasize
the role of the commensal gut microbiota in the development of the immune system, the
influence of the gut microbiota on immune development in infants is unclear.
Objective: To assess whether early colonization with certain gut microbiota species
associates with mucosal and systemic immune responses i.e. salivary SIgA and the
spontaneous toll like receptor (TLR) 2 and TLR4 mRNA expression and LPS-induced
cytokine/chemokine responses in peripheral blood mononuclear cells (PBMCs).
Methods: Faecal samples were collected at one week, one month and two months after birth
from 64 Swedish infants, followed prospectively to five years of age. Bacterial DNA was
analyzed with real-time PCR using primers binding to Clostridium difficile, four species of
bifidobacteria, two lactobacilli groups and Bacteroides fragilis. Saliva was collected at age
six and twelve months and at two and five years and SIgA was measured with ELISA. The
PBMCs, collected twelve months after birth, were analyzed for TLR2 and TLR4 mRNA
expression with real-time PCR. Further, the PBMCs were stimulated with LPS and
cytokine/chemokine responses were measured with Luminex.
Results: The number of Bifidobacterium species in the early faecal samples correlated
significantly with the total salivary SIgA levels at six months. Early colonization with
Bifidobacterium species, lactobacilli groups or C. difficile did not influence TLR2 and TLR4
expression in PBMCS. However, PBMCs from infants colonized early with high amounts of
Bacteroides fragilis expressed lower levels of TLR4 mRNA spontaneously. Furthermore,
LPS-induced production of inflammatory cytokines and chemokines, e.g. IL-6 and CCL4
(MIP-1β), were inversely correlated to the relative amounts of Bacteroides fragilis in the
early faecal samples.
Conclusion: Bifidobacterial diversity may enhance the maturation of the mucosal SIgA
system and early high colonization with Bacteroides fragilis might down-regulate LPS
responsiveness in infancy.
Gut microbiota, lactobacilli, bifidobacteria, Clostridium difficile, Bacteroides fragilis,
SIgA, TLR2, TLR4, infant
APRIL A proliferation-inducing factor
cDNA complementary deoxyribonucleic acid
ELISA Enzyme linked immunosorbent assay
G- Gram negative
G+ Gram positive
GALT Gut associated lymphoid tissue
MAMPs Microbial associated molecular pattern
mRNA messenger ribonucleic acid
PBMCs Peripheral blood mononuclear cells
PRR Pattern recognition receptor
PSA Polysaccharide A
RT PCR Reverse transcriptase polymerase chain reaction
SIgA Secretory IgA
Th T helper cell
TLR Toll-like receptor
TNF Tumour necrosis factor
TSLP Thymic stromal-derived lymphoprotein
A reduced microbial pressure in Westernized countries is postulated to underlie
the increase in allergy development during the past decades. Alterations in the early gut
microbiota may precede allergy development (1,2). Children developing allergy are,
compared to those who remain non-allergic, not as often colonized with bifidobacteria and
enterococci but more frequently colonized with clostridia including Clostridium (C.) difficile
early in life (1,2). Bifidobacterium colonization at species level might also be associated with
allergy (3-5). Furthermore, children who develop allergy during their five years of life were
already during the first week of life less often colonized with lactobacilli (L.) group I,
comprising L. rhamnosus, L. casei and L. paracasei, as compared to children not developing
allergy (3). However, not all studies demonstrate a relationship between the early gut
microbiota and allergy development (6). Whether pre- and postnatal administration of
probiotic bacteria associates with decreased incidence of allergic disease is unclear. Less IgE-
associated eczema (7) and less allergy development up to two years (8) but not five years of
age (9) are reported in probiotic-treated infants. However, others studies find no any
association between probiotic administration and allergy development (10). In order to
understand whether the early gut microbiota is associated with allergy development the
possible mechanisms, explaining how the early gut microbiota influence infant immune
responses and thus subsequent allergy development, need to be investigated.
Animal studies have emphasized the importance of the gut microbiota in
educating the immune system. The gut associated lymphoid tissue (GALT) in germ-free (GF)
animals is underdeveloped (11,12) with few IgA+ B cells (13). In addition, also the spleen of
GF mice contains fewer numbers of CD4+ T cells (14). Colonization with lactobacilli strains
increases the numbers of IgA+ B cells (13) and a polysaccharide from Bacteroides fragilis
(PSA) could restore the proportion of splenic CD4+ T cells (14). Also, PSA administration
increases IL-10 production (15). The early colonization appears to be of particular
importance as Bifidobacterium infantis could restore Th1 responses in neonatal but not in
adult ex-germ-free mice (16). Additionally, the serum levels of IgA appears to increase in
piglets colonized shortly after birth but remain low in GF piglets (17).
Studies in humans indicate that the early colonization with certain bacteria
influence systemic immune responses. For example, Bacteroides fragilis colonization in
infancy appears to increase the number of circulating IgA and IgM antibody producing cells
(18). Furthermore, infants who received a mixture of probiotic strains from birth had higher
plasma levels of C-reactive protein, total IgA, total IgE and IL-10 at six months than infants
in the placebo group (19). Although the composition of the gut microbiota at six months does
not appear to influence the salivary IgA levels at that age (20), it is conceivable that the
microbiota that colonizes the gut shortly after birth might influence immune development.
The innate immune compartment responds to different microbial associated
molecular patterns (MAMPs) by pattern recognition receptors (PRRs) expressed on immune
cells, e.g. dendritic cells, and mucosal epithelia (21). The capacity to respond to PRR signals
is important for adaptive immune responses such as IL-12 dependent direction of naïve T
cells into Th1 cells (22). The PRR Toll like receptor (TLR) 2 responds to MAMPs such as
lipoteichoic acid from Gram positive (G+) bacteria, while the TLR4 recognizes the endotoxin
lipopolysaccharide (LPS) from Gram negative (G-) bacteria together with a complex with
CD14 and MD-2 (23). Soluble CD14 appears to be higher in the plasma of infants early
colonized with Staphylococcus (S.) aureus than in the plasma of non-colonized infants (24).
However, it is not known how the early infant gut microbiota, consisting of both G+ and G-
bacteria, influences TLR responsiveness.
High salivary secretory IgA (SIgA) may protect sensitized children from
developing allergic symptoms and non-allergic children tend to have higher salivary SIgA
levels than allergic children (25). Recently, we also demonstrated that Swedish children have
less SIgA early in life compared to Estonian age-matched children (Tomicic et al
unpublished). As Estonian children are frequently colonized with lactobacilli (26), we
hypothesize that the early gut microbiota, notably lactobacilli and bifidobacteria at species
level, could influence the maturation of salivary SIgA production. The possible allergy-
protective effects from the increased pre- and postnatal microbial exposure in farming
environments might increase the expression of several PRRs e.g. TLR2 and TLR4 (27,28).
Thus, we also hypothesized that early exposure to certain intestinal microbes, the G-
Bacteroides fragilis and the G+ C. difficile, bifidobacteria and lactobacilli, modulates TLR
expression and LPS responsiveness. In addition, as Bacteroides fragilis colonization/PSA
administration induces IFNγ production in GF mice (14), we hypothesized that early
Bacteroides fragilis colonization influences spontaneous and PHA-induced IFNγ production
The birth cohort, comprising 123 Swedish children born between March 1996 and
October 1999, has been described in detail elsewhere (29). Briefly, the children were born at
term and had an uncomplicated perinatal period. Inclusion in the present study was based on
availability of faecal samples at one week, one month and/or at two months of age. In all, 64
infants were included. From the majority of these children blood samples collected at twelve
months, and/or salivary samples, collected at six months, twelve months, two years and/or
five years were also available. The study was approved by the Regional Ethics Committee for
Human Research at Linköping University. The parents of all children gave their informed
consent in writing. Clinical examinations of the children were made at three or six and twelve
months and at two and five years. At these occasions, skin prick tests were performed, and
questionnaires were completed regarding, in example, symptoms of allergy and use of
antibiotics. As development of allergic disease in relation to their early gut microbiota has
been investigated in a previous study in these children (3), this will not be discussed here.
A majority of the children had a history of atopic disease in the immediate family
(78%, Table 1). In total, three children were delivered with caesarean section. Most infants
were exclusively breastfed during their first three months (83%) and only two infants
received antibiotics during this time.
Analysis of bacterial DNA in the faecal samples
The analysis of the bacterial DNA has been described previously (3). In short,
faecal samples were collected into sterile plastic containers by the parents when the infants
were one week (collected at day five or six), one month and two months old. The samples
were stored at -70ºC until analysis.
Qiamp DNA Stool Mini Kit™ (Qiagen, Hilden, Germany) was used for the
isolation of DNA from 180-220 mg faeces and the included protocol for increasing the
bacterial DNA over human DNA was used. The concentration of nucleic acids was measured
with BIO-RAD Smartspec (Bio-Rad laboratories, Hercules, CA, USA) at 260 nm using BIO
RAD trUView Disposable Cuvettes (Bio-Rad laboratories, Hercules, CA, USA).
Bacterial DNA was analyzed with real-time PCR using primers binding to C.
difficile, B. adolescentis, B. longum, B. bifidum, B. breve, lactobacilli group I (comprising L.
rhamnosus, L. casei and L. paracasei), lactobacilli group II (comprising L. gasseri and L.
johnsonii) and Bacteroides fragilis. Sequences and concentrations of primers are described in
(3). The primers were used due to their specificity in binding to the specific bacterial DNA,
as well as for their suitability in SYBR Green PCR chemistry. Reference bacterial DNA, used
as standard and positive control, was purchased from LGC Standards (Borås, Sweden) and
BIOTECHON Diagnostics (Potsdam, Germany). The SYBR Green real-time PCR was
performed using 96-well detection plates in ABI prism 7000 (Applied Biosystem, Stockholm,
Sweden). All samples were performed in triplicates. Each well contained 2xPower SYBR
Green mastermix (Applied Biosystems, Stockholm, Sweden), forward and reverse primer
(MWG-Biotech, Edersburg, Germany), DNA and water. The Absolute Quantification
protocol in 7000 System software version 1.2.3f2 (Applied Biosystems) was employed and
the amplification was performed using the default program of 40 cycles, which also included
melting curve analysis. The software calculated the amount of specific bacterial DNA from
the standard curve, constructed from known amounts of reference bacterial DNA (5ng diluted
in 10-fold dilution series down to 50 fg). To avoid detecting false positives, triplicates with
CT values above 35 were considered as negative. The amount of the specific bacterial DNA
was then related to the total amount of nucleic acids in each sample. The specific bacterial
DNA is thus expressed as percent specific bacterial DNA of total nucleic acids and referred
to as relative amounts of specific bacterial DNA. The detection limit was 5*10-6 % specific
bacterial DNA of total nucleic acids. Negative samples were assigned a value ten times below
the detection limit i.e. 5*10-7 % specific bacterial DNA of total nucleic acids, and used in the
Analysis of total secretory IgA in saliva
Saliva was collected at six and twelve months and at two and five years of age
from the buccal cavity using a hand-pump connected to a thin plastic tube and immediately
frozen at -20oC. Before analysis of SIgA, the samples were heated at 56 oC for 30 minutes
and then centrifuged at 5000g for 15 minutes. Total SIgA was analyzed with ELISA using an
anti-human secretory component antibody (Dakopatts AB) as coating antibody, as previously
described (25). Human IgA (Sigma Immunochemicals) was diluted in seven steps for the
standard curve. The SIgA, bound to the coating antibodies, was detected with alkaline
phosphatase conjugated goat anti-human IgA antibodies (Sigma Immunochemicals) and
FAST pNPP substrate. The detection range was 16 to 1000 ng/mL for SIgA.
Venous blood samples were drawn into heparinized tubes (Becton Dickinson,
Stockholm, Sweden) at 12 months of age. Peripheral blood mononuclear cells (PBMCs) were
isolated on a Ficoll Paque density gradient (Pharmacia Biotech, Uppsala, Sweden). The cells
were thereafter cryopreserved according to standard methodology in 10% dimethyl sulfoxide
(Sigma-Aldrich, Stockholm, Sweden), 50% foetal calf serum and 40 % RPMI-1640 (Life
Technologies AB, Täby, Sweden).
Reverse transcription (RT) PCR of mRNA and quantification of gene expression
250 µL of cell suspension (1x106 viable cells/mL (as checked by Trypan blue
exclusion) in AIM-V serum free medium (Life Technologies AB) with 20 µM
mercaptoethanol (Sigma-Aldrich)) was cultivated for 24h in 37°C with 5% CO2 (Forma
CO2-incubator model 3862, Forma Scientific Inc., Marietta, Ohio, USA) with no added
stimulus. Thereafter the cells were lysed with RLT lysis buffer (Qiagen, Hilden, Germany).
The cell lysates were stored in -70° C until RNA isolation. Total RNA was isolated using
RNeasy™ 96 Protocol (Qiagen, Hilden, Germany) according to manufacturer´s instructions.
Briefly, the cells were lysed by RLT lysis buffer, mixed with ethanol and applied to
RNeasy™ 96 well plates. Contaminants were washed away by buffers and the RNA was
eluted in 2 x 30µl of RNase free water.
RNA was converted to complementary DNA (cDNA) using High Capacity
Archive Kit (Applied Biosystems, Foster City, CA, USA) according to manufacturer´s
instructions. Briefly, RNA was mixed with MultiScribe reverse transcriptase, random
primers, dNTPs, reverse transcription buffers and RNAse free water in 40µl reactions and run
for 10 min at 25° followed by 37°C for 120 min. The gene expression analysis was performed
with quantitative real-time PCR. Taqman® Gene Expression and Taqman® Fast Universal
PCR Master Mix were purchased from Applied Biosystems. The assay id:s were TLR2
Hs00610101_m1 and TLR4 Hs00152939_m1. Primers, probes, Master Mix, water and cDNA
was mixed and the samples were run on an Applied Biosystems 7500 Fast Real-Time PCR
system. The thermal cycling conditions were 95°C for 20s, followed by 40 cycles of 95°C
for 3s and 60°C for 30s. rRNA was used as internal controls, i.e. the amount of the expressed
gene was calculated relative to the amount of rRNA in each sample. Standards were used to
create a standard curve from which the amounts were calculated in each run using the
standard curve method as described in User Bulletin no 2 (Applied Biosystems). The inter-
assay variation was <6% for both genes and the slopes of the standard curves varied between
-3.3 to -3.9. Each sample was run in duplicates and a CV of maximum 15% was allowed.
Cytokine and chemokine analysis
After thawing, 0.25 mL of 1*106 viable cells/mL cell suspension in AIM-V serum
free medium were cultured in 37°C with 5% CO2. The cells were either cultured with 10
ng/mL LPS Salmonella enterica serotype thyphimurium (Sigma-Aldrich) or with medium
alone. The viability was checked by trypan blue exclusion. After 24h the supernatants were
collected and stored at -70°C until analysis. The cell supernatants were analysed with a
multiplex assay kit, according to the manufacturer’s instructions (Human cytokine 9-plex
panel, Bio-Rad Laboratories, Hercules, CA, USA). The assay detects the following analytes;
IL-6, CXCL8 (IL-8), IL-10, IL12p70, IL-17, IL-1β, CCL2 (MCP-1), CCL4 (MIP-1β), and
TNF. The samples were analysed on a Luminex100 instrument (Biosource, Nivelles, Belgium)
and the data was analysed with the software StarStation 2.0 (Applied Cytometry Systems,
Sheffield, UK). The lower detection limit was 10 pg/mL for IL-6 and CCL4, 6 pg/mL for IL-
10, CXCL8 and TNF, 2 pg/mL for IL12p70, 3 pg/mL for IL-17, 4 pg/mL for IL-1β, and 18
pg/mL for CCL2. Comparisons of LPS-induced cytokine and chemokine responses were
made after the control value, i.e. responses from cells cultured in medium alone, was
For IFNγ analyses, 1 mL of cell suspension (1*106 viable cells/mL in AIM-V
serum free medium supplemented with 20µM mercaptoetanol (Sigma-Aldrich)) was grown in
duplicates alone or with 2µg/mL phytohaemagglutamin (PHA) (Sigma-Aldrich). The culture
conditions and analyses have been described in more detail a previous paper (30). After one
and six days of culture, the supernatants were collected and stored at -20ºC. Interferon γ was
measured with an ELISA kit (CLB Pelikine CompactTM, Research Diagnostics Inc., Flandern,
NJ, USA). The detection limit was 25 pg/mL. Spontaneous IFNγ production was detected
after six days, but not one day of culture.
Spearman’s rank coefficient was calculated to investigate whether relative
amount of bacterial DNA and/or number of Bifidobacterium species in faecal samples
correlated with concentrations of total salivary SIgA and the relative mRNA expression of
TLR2 and TLR4. It was also calculated to understand whether the bacterial amounts
influenced the production of cytokines and chemokines from PBMCs after microbial
stimulation. Mann-Whitney U test was performed to evaluate whether infants colonized with
the specific bacteria had different levels of total SIgA in saliva compared with un-colonized
infants. In addition, Mann-Whitney U test was calculated to understand whether the
colonized and un-colonized groups expressed different levels of TLR2 and TLR4 mRNA and
produced different levels of IFNγ at 12 months. Fisher´s exact test was calculated to
understand whether colonization with Bacteroides fragilis was associated with spontaneous
IFNγ production. Many statistical tests were performed and thus p-values close to 0.05 might
be false significances. As this study was not sufficiently powered to detect very low p-values,
we only report significant values observed at several time points, or significant values
observed at one occasion with at least a tendency at another time point. This approach would
decrease the risk of including false significances. The study should be viewed as exploratory
and consequently p<0.05 were chosen as statistically significant.
Sjögren 14 Download full-text
Gut microbiota during the first two months
Bifidobacteria and Bacteroides fragilis were commonly present already in one-
week old infants, whereas the other bacteria tended to become more frequently detected as
the infants grew older (table 2). Lactobacilli occurred in lower amounts than bifidobacteria
and Bacteroides fragilis. Few infants were colonised with C. difficile.
Total salivary secretory IgA associates with the early gut microbiota.
The number of Bifidobacterium species in faeces collected one week, one month
and two months after birth correlated with total salivary SIgA at six months (r=0.51 to 0.58,
p=0.02 to 0.045, Fig. 1 exemplifies the one month colonization) but not in older children.
When analyzing the different Bifidobacterium species separately, it was shown that infants
colonized with B. adolescentis at one and two months had significantly higher levels of SIgA
at six months (Fig. 2a and median; 12.0 (4.3-19-7) µg/mL, respectively) compared to non-
colonized infants (Fig. 2a and median; 5.3 (2.0-10.0) µg/mL, respectively). Furthermore, the
intensity of one and two month B. adolescentis colonization (expressed as percentage B.
adolescentis DNA of all faecal nucleic acids), was associated with higher SIgA at six months
(r=0.66 and 0.55, p=0.007 and 0.03, respectively). In addition, SIgA levels at six months
tended to be higher in infants colonized with B. breve at one week and one month (median;
9.0 (6.2-17.0) µg/mL and Fig 2a, respectively) than in non-colonized infants (median; 4.3
(2.0-15.5) µg/mL, p=0.02 and Fig 2a, respectively). The intensity of B. bifidum colonization
at one week, one month and two months after birth was correlated with the SIgA production
at twelve months (r=0.41 to 0.47, p=0.01 to 0.04, Fig. 3 exemplifies one month colonization).
The levels of SIgA at six months and five years were significantly associated with
the colonisation with lactobacilli group I at one month after birth (Fig. 2a and d). Also,